Pandas 3.0
16 days ago
- #data-science
- #python
- #pandas
- Pandas 3.0.0 has been released with significant improvements and some breaking changes.
- Highlights include a dedicated string data type by default for better performance and type safety.
- Copy-on-Write (CoW) behavior is now the default, making operations more predictable and improving performance.
- New default resolution for datetime-like data avoids out-of-bounds errors for certain dates.
- Introduction of `pd.col` syntax for simplified callable creation in `DataFrame.assign`.
- Recommendation to upgrade to Pandas 2.3 first to ensure code compatibility before moving to 3.0.
- Breaking changes include the new string dtype and changes in copy/view behavior, requiring code updates.
- Installation instructions provided for both PyPI and conda-forge.
- Users are encouraged to report issues on the pandas issue tracker.